Carbon Monoxide and Nitric Oxide Mediate Cytoskeletal Reorganization in Microvascular Cells via Vasodilator-Stimulated Phosphoprotein Phosphorylation: Evidence for Blunted Responsiveness in Diabetes

Human lung microvascular endothelial cells (HMECs) (LONZA, Walkersville, MD) were cultured to 80% confluency in EGM-2 MV medium (LONZA) in 5.5 or 25 mmol/l glucose for 1 week. Cells were then incubated overnight in EBM-2 supplemented with 2% fetal bovine serum. Fresh medium (EBM-2 + 2% fetal bovine serum) was added on the following day, and CO- or NO-containing medium was added to the designated samples for 1 or 15 min at 37°C, as described below.

Peripheral blood from healthy and diabetic volunteers giving informed consent under a protocol approved by the institutional review board was collected in cell preparation tubes (CPTs) with heparin (BD Biosciences, San Jose, CA), and platelets were isolated using Opti-Prep (Axis-Shield PoC, Oslo, Norway), as recommended by the manufacturer.

Peripheral blood from healthy and diabetic volunteers was then collected into CPTs (BD Biosciences), and the EPCs were freed of the mononuclear cells using an Endocult medium kit (Stem Cell Technologies, Vancouver, CA) according to the manufacturer's protocol, as previously described (13).

EPCs were cultured onto fibronectin-coated dishes with Endocult stem cell liquid media until colonies formed. These cells were then treated for 15 min or 4 h with 100 μmol/l diethylenetriamine (DETA)-NO (Sigma-Aldrich, St. Louis, MO) in water or 10 μmol/l CO-donor Ru(II)(CO)₃Cl₂ (Sigma-Aldrich) in DMSO. Control samples were treated with vehicle. At the end of the treatment, medium was removed and cells were fixed overnight at 4°C in 4% paraformaldehyde in PBS, supplemented with calcium and magnesium ions, and adjusted to pH 7.4.

Cells were then washed in PBS and permeabilized with 0.1% Triton X-100 for 30 min at room temperature. After washing the samples an additional three times in PBS, cells were treated with 10% normal goat serum–1% BSA in PBS for 1 h at room temperature to block nonspecific antigens. Cells were then incubated with 5 μg/ml mouse anti-VASP antibody (BD Biosciences) in 5% normal goat serum overnight at 4°C and then with fluorescein isothiocyanate (FITC)-labeled goat anti-mouse IgG₁ (1:250 dilution) (Southern Biotech, Birmingham, AL) in 5% normal goat serum for 1 h at room temperature. Samples were then washed, dried, and mounted with 4′,6-diamino-2-phenylindole (DAPI) (Vectashield; Vector Laboratories, Burlingame, CA) to label nuclear DNA. Samples were examined by fluorescence microscopy (Nikon Eclipse TE200) using a Nikon Plan-fluor ×40 (NA = 1.30) oil objective and a fluorescein filter set (520 ± 2 nm). Micrographs were captured using a SPOT digital camera 0.60X HRD06-NIK (Diagnostic Instruments, Sterling Heights, MI) and processed using SPOT Advanced software Version 2.2.1 for Windows.

Peripheral blood was collected into CPTs (BD Biosciences) and centrifuged to obtain the mononuclear cell fraction, from which CD34⁺ cells were isolated using a CD34⁺ isolation kit (Stem Cell Technologies, Vancouver CA). Briefly, 2 × 10⁻⁷ cells were incubated with a CD34⁺ selection cocktail for 15 min, and 50 μl magnetic nanoparticles was then added and incubated for another 10 min. The suspension volume was increased to 2.5 ml, and the particle-bound CD34⁺ cells were concentrated on the tube's inner surface with the aid of a magnet. The supernatant was decanted, and the remaining CD34⁺ cells were resuspended in RPMI culture media (Cellgro, Herndon, VA).

The cells were then treated with 100 μmol/l on DETA-NO (Sigma-Aldrich) for either 15 min or 4 h in a CO₂ incubator. Following treatment, the CD34⁺ cells were permeabilized using a cytofix/Cytoperm kit (BD Biosciences) and blocked with 10% normal human serum (Jackson Immuno Research labs, West Grove, PA), and 10 μg antiphospho-VASP (Upstate Signaling, Lake Placid, NY) was added to the cells, followed by 30-min incubation on ice. The cells were then washed with PBS and incubated with 23 μg FITC-conjugated goat anti-mouse (Jackson Immuno Research labs, West Grove, PA), and again incubated on ice for 30 min. The resulting cells were washed in PBS and analyzed by fluorescence-activated cell sorting (FACS). Anti-GFP (Molecular probes, Carlsbad, CA) was used as an isotype control.

The half-lives for NO and CO release from DETA-NO and Ru(II)(CO)₃Cl₂ are greatly different, resulting in the almost immediate release of CO but very slow release of NO. Therefore, to assure that cells were exposed to comparable concentrations of NO and CO, we first calculated the amount of CO released at 1 and 15 min using the integrated rate law: [CO donor]_t/[CO donor]_initial = exp (−k_COt), where [CO donor]_t and [CO donor]_initial are the CO donor concentrations at time t and, initially, k_CO is the first-order rate constant for CO release, and t is the incubation period. We then used the corresponding rate law, i.e., [NO donor]_t/[NO donor]_initial = exp (−k_NOt) to determine the time needed to obtain an extent of NO release comparable with that observed with CO donor in 1 and 15 min. The CO and NO donor rate constants k_CO and k_NO of 0.07 min − 1 and 0.0006 min − 1, respectively, were determined based on the published half-lives of 10 min for Ru(II)(CO)₃Cl₂ (18) and 18 h for DETA-NO (19). In our experiments, we therefore separately preincubated the NO donor in culture medium in a sealed tube for the calculated NO times, and then transferred an appropriate aliquot to the cell. Because the partial pressure exerted by the released NO and CO is far below the solubilities in aqueous solution, and because the 1- and 15-min exposure times were short, there was no need to maintain treated cells in a sealed culture dish.

Following treatment with either the CO or NO donor, the CD34⁺ cells or platelets were washed with PBS and lysed with lysis buffer (Cell Signaling, Danvers, MA) containing protease (catalog no. P8340l Sigma-Aldrich) and phosphatase inhibitors (catalog no. P2850; Sigma-Aldrich). Protein was quantified in each sample using a commercially available BCA kit (Pierce, Rockford, IL), and 50 μg was applied to each well above the polyacrylamide gel (7.5%) (Criterion; Biorad Laboratories, Richmond, CA) at 120 V for 20 min, followed by 140 V for 65 min. Samples were run in duplicate for later detection of the two phosphorylated VASP (pVASP) isoforms. The separated proteins were transferred to a nitrocellulose membrane (Biorad Laboratories) in a semidry transblot apparatus (Biorad Laboratories). The membranes were blocked in Tris-buffered saline (TBS) containing 5% milk and 0.05% Tween-20 (Sigma-Aldrich) for 1 h at room temperature. For the detection of pVASP isoforms, the membranes were incubated at 4°C overnight with either a mouse monoclonal anti-pVASP Ser-239 antibody (Upstate Cell Signaling Solutions, Lake Placid, NY) or a rabbit polyclonal anti-pVASP Ser-157 antibody (Santa Cruz Biotechnology, Santa Cruz, CA), both at a 1:200 dilution. Blots were then washed with 5% milk for 5 min and incubated with a 1:2,000 dilution of either a horseradish peroxidase (HRP)-conjugated anti-mouse antibody (Sigma-Aldrich) or an anti-rabbit (Santa Cruz Biotechnology) for 1 h at room temperature. After incubation with the secondary antibody, the membranes were washed twice with TBS containing 0.05% Tween-20 for 5 min, followed by two additional 10-min washes with TBS. Protein bands were visualized with enhanced chemiluminescence using a commercial Western blot detection kit (Amersham Biosciences, Amersham, U.K.) and pVASP levels were quantified using Scionimage software (Scion, Frederick, MD). Standard molecular weight markers (Biorad Laboratories) served to verify the molecular mass of pVASP (50 kDa).

After pVASP detection, the membranes were reprobed for VASP using a 1:200 dilution of a goat anti-VASP antibody (Santa Cruz Biotechnology) and for β-actin using a mouse anti–β-actin primary antibody at a dilution of 1:7,000 (Sigma-Aldrich), followed by HRP-conjugated anti-goat (Santa Cruz Biotechnology) and anti-mouse IgG antibody (Sigma-Aldrich), both at 1:2,000 dilution, and were visualized by enhanced chemiluminescence. Standard molecular weight markers served to verify the 42-kDa molecular mass of β-actin.

CD34⁺ cells were cultured in defined serum-free medium (StemSpan SFEM; StemCell Technologies, Vancouver, Canada) to obtain an optimal number of cells. For CD34⁺ cell proliferation and expansion, without differentiation, we used 1 ml StemSpan SFEM supplemented with a cytokine cocktail (100 ng/ml flt3 ligand, 100 ng/ml stem cell factor, 20 ng/ml interleukin-3, and 20 ng/ml interleukin-6; StemCell Technologies) and 50 ng/ml thrombopoietin (R&D Systems, Minneapolis, MN). After determining cell number with a hemocytometer (Hausser Scientific, Horsham, PA), 10,000 cells were transferred to each well in a 96-well plate (BD Falcon, San Jose, CA). After incubation with/without CO or NO donor, cells were fixed with 4% paraformaldehyde for 20 min at room temperature and then harvested by centrifugation. After removing the fixation solution, cells were permeabilized with 0.1% Triton X-100 in PBS and blocked with LI-COR Odyssey blocking buffer (LI-COR Biosciences, Lincoln, NE) for 1.5 h at room temperature with moderate shaking. Blocking buffer was removed by aspiration, and the cells were incubated with 50-μl volume of diluted primary antibody at 4°C overnight in a cold room. Cells were then washed four times for 5 min in PBS containing 0.1% Tween-20 (Fisher Scientific, Pittsburgh, PA) and incubated with diluted secondary antibody. After 1-h incubation, cells were washed four times (5 min each) with PBS containing 0.1% Tween-20, followed by final washing in PBS to remove excessive detergent. The 96-well plate was then scanned in the appropriate channels using an Odyssey infrared imaging system (LI-COR Biosciences). Relative quantification was normalized and readjusted in cell number from well to well using DNA staining.

All reported data are expressed as the mean ± SEM and were obtained from at least three independent experiments. Student's t test and ANOVA with Student-Newman-Keuls posttest were used to compare the data.

The choice of the technique used (e.g., standard Western blotting, In-Cell Western analysis, and FACS) for these studies was largely governed by the number of cells available for analysis. Because CD34⁺ cells represented <1% of circulating mononuclear cells, In-Cell Western (low sensitivity) and FACS (higher sensitivity) analyses were the methods of choice; however, for more abundant cells such as endothelial cells, standard Western blotting was sufficient. All the phosphorylation studies on EPCs were performed after 15 min of treatment. We then tested whether a shorter treatment (1 min) could still elicit a significant migratory response.

We therefore pretreated human CD34⁺ EPCs for 1 and 15 min with medium containing NO or freshly generated CO, followed by SDF-1 stimulation (13), and then assessed cell migration using Boyden chambers. As demonstrated in Fig. 1, both CO and NO treatment resulted in increased EPC migration to SDF-1, although a more robust response was observed with NO at 1 min.

Because VASP is a key component of the actoclampin molecular motors responsible for actin-based cell motility (14,20), and because VASP was originally identified as a major substrate for cGMP-stimulated protein kinase in platelets and endothelial cells (21), we investigated whether NO and CO elicited similar effects on VASP phosphorylation and its mobilization to the leading edge of motile cells. VASP function appears to be chiefly regulated by phosphorylation at Ser-157 and Ser-239, and the availability of phosphorylation site-specific antibodies provides an unambiguous way to define both the sites and extents of signal molecule–activated VASP phosphorylation. We therefore incubated CD34⁺ cells for 15 min in the presence of NO or CO already liberated from their respective donors. As shown in Fig. 2, NO treatment routinely increased the extent of VASP phosphorylation at Ser-239, whereas CO treatment consistently increased the extent of Ser-157 phosphorylation in these cells.

To confirm that the differential effects of CO and NO are not limited to CD34⁺ cells, we also examined platelets, which are a rich source of VASP and are the cells wherein VASP phosphorylation is best characterized (22). In platelets exposed to either NO or CO for 1- and 15-min periods, we found that CO treatment caused a modest yet significant VASP phosphorylation at Ser-157, whereas NO treatment significantly increased VASP phosphorylation at Ser-157 with maximum response at 1 min and at Ser-239 with maximum effect at 15 min (Fig. 3). These findings suggest that CO and NO elicit markedly different VASP phosphorylation responses in platelets and CD34⁺ cells.

Because VASP resides on the cytoplasmic face of the leading edge in motile cells (23), and because VASP phosphorylation is believed to control its interactions with VASP-docking proteins like vinculin, zyxin, and migfilin, we next examined the effects of donor-generated NO and CO on the redistribution of VASP in microvascular endothelial cells growing while firmly adhered to fibronectin-coated culture dishes. Under basal conditions (i.e., no NO or CO), VASP was mainly localized along the sides of actin filaments found throughout the peripheral cytoplasm (Fig. 4A and B). However, following a 15-min exposure to CO, VASP redistributed to the leading edge of advancing microvascular endothelial cells (Fig. 4C and D). VASP redistribution was also observed when these cells were incubated with NO (Fig. 4E and F).

In as much as diabetes-associated vascular dysfunction is frequently attributed, at least in part, to reduced levels of bioavailable NO, we sought to determine whether diabetes impacted the NO and CO effects on VASP phosphorylation within cell types typically affected by diabetes. We treated platelet samples from diabetic individuals with NO or CO under the same conditions used in Fig. 1 and observed no change in VASP phosphorylation, at either Ser-157 or Ser-239 (data not shown). In Fig. 5, VASP phosphorylation is shown for two diabetic patients and one nondiabetic control subject. Patient 1 had type 2 diabetes of 5 years duration and excellent glycemic control (A1C 6.4%), while patient 2 had type 1 diabetes of 48 years duration and poor glycemic control (A1C 11.3%). A different pattern was observed in the diabetic CD34⁺ cells, where NO treatment stimulated phosphorylation at VASP's Ser-239 to a significantly less extent, when compared with the nondiabetic control cells. However, we observed considerable patient-to-patient variation in the degree of the response but all with the same pattern.

Notably, compared with control cells grown in low-glucose medium, exposure of microvascular cells to conditions mimicking aspects of diabetes (e.g., culturing these cells for 1 week in high-glucose medium) resulted in a failure of NO and CO to elicit changes in either VASP redistribution to the leading edge of the cells, as shown by immunofluorescence (compare Figs. 4 and 6), or phosphorylation at Ser-157 or Ser-239, as detected by standard Western blot analysis (Fig. 7).

Understanding the regulation of endothelial cell migration in response to chemokines has proven to be a daunting task that requires the investigation of a manifold of factors known to affect chemotaxis (e.g., chemokine sensing by receptor-mediated signaling, cellular locomotion by actin-based molecular motors, and even the cell's energy status). Previous studies from our laboratory on CD34⁺ EPCs showed that diabetic EPCs have reduced intracellular NO concentration as well as a concomitantly reduced migratory capability. We found that, when pretreated with an NO donor, cell migration can be restored (13), and we further demonstrated that this exogenous NO exposure resulted in enhanced diabetic cell migration, with attendant changes in phosphorylation of the actin cytoskeletal protein VASP (13).

Given the robust nature of the vasodilator gases CO and NO in modulating vascular function, we focused on the following questions: Can CO suffice in place of NO in promoting vascular cell migration? If so, do NO and CO exert their signaling effects on the same downstream phosphorylation target(s)? In this study, we demonstrate that CO can regulate VASP phosphorylation and in turn alter cell migration. We also directly compare the CO effects to those of NO in EPCs, platelets, and microvascular endothelial cells. Our studies demonstrate that both vasoactive gases promote cytoskeletal changes through site- and cell type–specific VASP phosphorylation; however, these responses to NO and CO are blunted in diabetes and may be responsible for reduced vascular repair and tissue perfusion (24,25).

The actoclampins are the force-generating motors responsible for actin-based cell crawling and vesicle motility (14). Each of these membrane surface–bound molecular motors consists of an actin filament plus-end tracking protein called a clampin (e.g., VASP, N-WASP [neural Wiskott–Aldrich syndrome protein], formins, etc.) and its actin filament partner. The energy for force production appears to be derived from nucleotide hydrolysis at the filament's penultimate actin-ATP subunits, thereby promoting clampin release, translocation, and rebinding to terminal actin-ATP subunits. Before the active motor complex is assembled, clampins must be recruited to the membrane surface, where they dock at motility sites at the tips of filopodia and lamellipodia as well as in the focal adherens complex. Rottner et al. (23), for example, showed that VASP not only colocalizes to adhesion sites with the adaptor proteins vinculin and zyxin, but is also recruited to the tips of lamellipodia in amounts that are directly proportional to the rate of lamellipodial protrusion. Tokuo and Ikebe (26) further showed that myosin X specifically transports VASP and other members of the Ena/VASP clampin protein family to the leading edge, where VASP then binds to a membrane docking protein such as vinculin, zyxin, or migfilin (27,28). Only then can the actoclampin motor assemble and generate the forces needed for cell migration. When viewed from this perspective, VASP phosphorylation may affect the recruitment by myosin X, VASP docking with membrane components, and/or VASP-mediated formation of an active motor that can propel cell crawling. Moreover, as cells stop moving, VASP is known to redistribute to other intracellular sites. Benz et al. (29), for example, showed that VASP interacts with αII-spectrin in endothelial cells and that PKA-mediated VASP phosphorylation at Ser-157 inhibits this binding interaction. They also showed that VASP is dephosphorylated upon formation of cell-cell contacts and that, in confluent cells, αII-spectrin colocalizes with nonphosphorylated VASP at cell-cell junctions (29). The exact details of how VASP phosphorylation at Ser-157 or Ser-239 controls one or more of these steps remain to be worked out. Even so, the observation that different cell types contain various clampins and their respective membrane-docking proteins is likely to explain why VASP phosphorylation can either stimulate or suppress cell crawling and actin-based cell shape changes in a manner depending on cell type and/or culture conditions.

VASP is also the most abundant platelet and endothelial protein phosphorylated by PKG in NO signaling pathways, and, as noted above, this cytoskeletal adaptor protein is also a PKA substrate. Three serine/threonine phosphorylation sites within VASP play roles resulting in the inhibition of platelet aggregation and focal adhesion assembly (30,31). In this study, we demonstrated that CO can also regulate VASP phosphorylation, and we directly compare CO and NO effects in EPCs, platelets, and microvascular endothelial cells. We found that CO pretreatment can similarly stimulate EPC migration (Fig. 1) and phosphorylation of VASP's Ser-157, whereas NO exposure results in Ser-239 phosphorylation.

While we did not perform direct measurements of CO levels in our experiments, we did use CO concentrations that are achievable in vivo in physiological conditions and previously used by investigators (32,33). Endogenous CO has been reported to be generated in many cell types, and the amount of CO released via the heme oxygenase reaction can reach up to 12 ml/day (∼16 μmol · l⁻¹ · h⁻¹) (32). Previous studies have reported that tissues can produce 0.1–100 μmol/l CO in vivo from the HO reaction (33); it is therefore quite reasonable to believe that sufficient levels of HO-1 are present in cells to provide sufficient levels of CO, explaining our observed changes in VASP phosphorylation.

Moreover, in response to incubation of microvascular endothelial cells with NO or CO donors, VASP was readily redistributed to the peripheral membrane and filipodia (Fig. 4). Thus, with regard to VASP localization, our data suggest that CO and NO may both have critical roles in vascular cell dynamics, and CO may have important contributions at low NO bioavailability, a condition already known to occur in diabetes. Our studies clearly indicate that both vasoactive gases promote cytoskeletal changes through site- and cell type-specific VASP phosphorylation and that responses to NO and CO are blunted in diabetes. While the migratory deficiency seen in diabetic EPCs could be overcome by exogenous NO (13), exogenous NO did not result in phosphorylation and redistribution of VASP in mature endothelial cells cultured in high glucose conditions. This defect can be viewed as “diabetes-induced NO resistance.”

What also becomes apparent from the present study is that exposure of cells to NO or CO can greatly alter VASP recruitment to the leading edge with consequential effects on cell motility. Perhaps even more significant, in the context of diabetes, is that culturing endothelial cells at high glucose conditions that mimic diabetes results in motility defects that can be traced, at least in part, to altered VASP phosphorylation. We therefore postulate that these defects contribute to reduced vascular repair and tissue perfusion.

Both heme oxygenases generate CO, but they do so with very different kinetics (34,35); HO-1 is induced by oxidants such as hydrogen peroxide, UV radiation, and proinflammatory cytokines and by growth factors, hemodynamic or shear stress, heat shock, and even by NO (3). Endothelial cells derived from either the micro- or macrovasculature (36) responded equally to CO and NO, reflecting the key roles of both HO-1 and endothelial NO synthase throughout in the entire vasculature. In contrast, platelets do not respond to CO, perhaps reflective of the low levels of heme oxygenase in adult platelets (37). By altering HO-1 and NOS gene expression, hypoxia potentially modulates the availability of these gaseous second messengers. Thus, fluxes in the CO and NO generation during hypoxic stress are likely to have dramatic consequences on the regulation of vascular functions such as dilation, expression of vasomodulators, inhibition of platelet aggregation, and smooth muscle cell proliferation (38). Although the findings in this report support a role for CO sufficing for NO in promoting vascular cell migration, the signaling action of CO results in the phosphorylation of different VASP sites than NO. Furthermore, we demonstrate that, while both vasoactive gases promote cytoskeletal changes through site- and cell type–specific VASP phosphorylation, these responses are blunted in diabetes and may be responsible for the altered vascular repair and tissue perfusion associated with diabetic vascular complications.

Support for this research came from National Institutes of Health Grants NIH1R01 EY07739 and NIHR01 EY12601 (to M.B.G.); the Juvenile Diabetes Research Foundation (to M.B.G.); and the American Heart Association (to S.L.C.).

We acknowledge Dr. Lynn Shaw for help in preparing the manuscript.